ggplot2中stat_summary_hex热力图未按均值正确缩放问题

尝试用ggplot2的stat_summary_hex绘制热力图，怀疑未按Prediction列的均值填充六边形，且热力图无法正确缩放。示例代码如下：

FFxBARHH1 <- data.frame("plate_x" = c("-0.20", "0.21", "-0.19", 
                                      "0.66", "0.10", "0.52"),
                  "plate_z" = c("3.18", "2.01", "2.56", "2.81", "1.93", 
                            "1.85"),
                  "Prediction" = c("0.916", "0.908", "0.872", "0.000", "0.000", 
                                "0.000"))

FFxBARHH1$Prediction<-  as.numeric(FFxBARHH1$Prediction)
FFxBARHH1$plate_x<-  as.numeric(FFxBARHH1$plate_x)
FFxBARHH1$plate_z<-  as.numeric(FFxBARHH1$plate_z)

ggplot(FFxBARHH1, aes(x = plate_x, y = plate_z, z = Prediction))+
stat_summary_hex(fun = mean, bins = 9) +
stat_summary_hex(geom = "text", aes(label = after_stat(value)), fun = length,
               bins = 9) +
scale_fill_gradient2(low = "blue", mid = "white", high = "red", 
                   midpoint = 0.240, 
                   limits = c(0.0, 1.0)) +
xlim(c(2,-2)) +
ylim(c(0,5)) +
coord_fixed(0.8)+
labs(title = "Gerrit Cole Fastball April 2022 ExAVG Map RHH")+
annotate('rect', xmin = -0.85, xmax = 0.85, ymin = 1.6, ymax = 3.5, 
       fill = 'black', color = 'black', alpha = 0.0001, size = 1)

你的代码核心逻辑正确，主要是两个细节问题导致可视化异常：

1. 修正x轴范围的顺序 #

xlim(c(2,-2))会反转x轴方向，打乱六边形的布局逻辑，导致缩放和位置异常。需将x轴范围改为下限在前、上限在后：

xlim(c(-2, 2))

2. 优化六边形数量适配小样本 #

当前stat_summary_hex(fun = mean)已经在计算每个六边形内Prediction的均值，但你的样本量仅6个数据点，bins=9会生成大量空六边形，导致热力图效果不明显。可以减少六边形数量，让现有数据点集中到少数六边形中，更直观看到均值填充效果：

stat_summary_hex(fun = mean, bins = 3)

修正后的完整代码 #

FFxBARHH1 <- data.frame("plate_x" = c("-0.20", "0.21", "-0.19", 
                                      "0.66", "0.10", "0.52"),
                  "plate_z" = c("3.18", "2.01", "2.56", "2.81", "1.93", 
                            "1.85"),
                  "Prediction" = c("0.916", "0.908", "0.872", "0.000", "0.000", 
                                "0.000"))

FFxBARHH1$Prediction<-  as.numeric(FFxBARHH1$Prediction)
FFxBARHH1$plate_x<-  as.numeric(FFxBARHH1$plate_x)
FFxBARHH1$plate_z<-  as.numeric(FFxBARHH1$plate_z)

ggplot(FFxBARHH1, aes(x = plate_x, y = plate_z, z = Prediction))+
  stat_summary_hex(fun = mean, bins = 3) +
  stat_summary_hex(geom = "text", aes(label = after_stat(value)), fun = length,
                   bins = 3) +
  scale_fill_gradient2(low = "blue", mid = "white", high = "red", 
                       midpoint = 0.240, 
                       limits = c(0.0, 1.0)) +
  xlim(c(-2, 2)) +
  ylim(c(0,5)) +
  coord_fixed(0.8)+
  labs(title = "Gerrit Cole Fastball April 2022 ExAVG Map RHH")+
  annotate('rect', xmin = -0.85, xmax = 0.85, ymin = 1.6, ymax = 3.5, 
           fill = 'black', color = 'black', alpha = 0.0001, size = 1)

额外说明 #

• 若想让文本标签显示均值而非数据点数量，只需将fun = length改为fun = mean，并可格式化标签（例如label = round(after_stat(value), 3)）；
• coord_fixed(0.8)是固定x、y轴比例，可根据可视化需求调整，不影响均值计算逻辑。

内容的提问来源于stack exchange，提问作者user19422662